{{ message }}
-
Notifications
You must be signed in to change notification settings - Fork 148
Expand file tree
/
Copy pathadjust_allocation.cpp
More file actions
584 lines (512 loc) · 21.7 KB
/
Copy pathadjust_allocation.cpp
File metadata and controls
584 lines (512 loc) · 21.7 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
/*
* The MIT License (MIT)
*
* Copyright (c) 2015-2026 Advanced Micro Devices, Inc. All rights reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include <migraphx/adjust_allocation.hpp>
#include <migraphx/dead_code_elimination.hpp>
#include <migraphx/pass_manager.hpp>
#include <migraphx/instruction.hpp>
#include <migraphx/make_op.hpp>
#include <migraphx/register_op.hpp>
#include <migraphx/module.hpp>
#include <test.hpp>
// Test allocation operation
struct test_allocate
{
migraphx::shape s;
template <class Self, class F>
static auto reflect(Self& self, F f)
{
return migraphx::pack(f(self.s, "shape"));
}
std::string name() const { return "test::allocate"; }
migraphx::shape compute_shape(const std::vector<migraphx::shape>&) const { return s; }
};
MIGRAPHX_REGISTER_OP(test_allocate);
// Test copy operation
struct test_copy
{
std::string name() const { return "test::copy"; }
migraphx::shape compute_shape(const std::vector<migraphx::shape>& inputs) const
{
return inputs.at(1);
}
// Not context-free: takes context parameter
migraphx::argument
compute(migraphx::context&, const migraphx::shape&, std::vector<migraphx::argument> args) const
{
return args.at(1);
}
std::vector<std::size_t> output_alias(const std::vector<migraphx::shape>&) const { return {1}; }
};
MIGRAPHX_REGISTER_OP(test_copy);
struct test_fill
{
std::string name() const { return "test::fill"; }
migraphx::shape compute_shape(std::vector<migraphx::shape> inputs) const
{
return inputs.front();
}
std::vector<std::size_t> output_alias(const std::vector<migraphx::shape>&) const { return {0}; }
};
MIGRAPHX_REGISTER_OP(test_fill);
// Test allocation model
struct test_allocation_model
{
std::string name() const { return "test::allocate"; }
std::string copy() const { return "test::copy"; }
migraphx::operation allocate(const migraphx::shape& s) const { return test_allocate{s}; }
migraphx::operation preallocate(const migraphx::shape& s, const std::string&) const
{
return test_allocate{s};
}
bool needs_out_params() const { return false; }
};
// Test operator that takes an output buffer but returns a specific shape
// regardless of the output buffer size. This is used to test that adjust_allocation
// will reallocate when the shapes don't match.
struct simple_op
{
migraphx::shape output_shape;
template <class Self, class F>
static auto reflect(Self& self, F f)
{
return migraphx::pack(f(self.output_shape, "output_shape"));
}
std::string name() const { return "simple_op"; }
migraphx::shape compute_shape(const std::vector<migraphx::shape>&) const
{
return output_shape;
}
// Not context-free: takes context parameter
migraphx::argument
compute(migraphx::context&, const migraphx::shape&, std::vector<migraphx::argument> args) const
{
return args.back();
}
// Output aliases the last input (the output buffer)
std::vector<std::size_t> output_alias(const std::vector<migraphx::shape>& inputs) const
{
return {inputs.size() - 1};
}
};
static void run_pass(migraphx::module& m)
{
migraphx::run_passes(
m,
{migraphx::adjust_allocation{test_allocation_model{}}, migraphx::dead_code_elimination{}});
}
// Test that adjust_allocation reallocates when the output shape differs from the allocated shape
TEST_CASE(realloc_shape_mismatch)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
// Allocate a buffer with shape {3, 2} but the operator returns shape {2, 3}
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {3, 2}}});
m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, alloc);
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
// After adjust_allocation, the allocate should have the correct shape {2, 3}
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
m2.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, alloc);
}
EXPECT(m1.sort() == m2.sort());
}
// Test that adjust_allocation does nothing when shapes already match
TEST_CASE(no_realloc_shape_match)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
// Allocate a buffer with the same shape as the operator output
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, alloc);
}
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
// Test that adjust_allocation skips when output alias is a parameter with matching shape
TEST_CASE(skip_output_param_shape_match)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
// Output parameter with same shape as what the operator produces
auto out = m1.add_parameter("output", {migraphx::shape::float_type, {2, 3}});
auto r = m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, out);
m1.add_return({r});
}
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
// Test that context-free operations are skipped
TEST_CASE(skip_context_free_op)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
// Use a context-free operation (add is context-free)
auto sum = m1.add_instruction(migraphx::make_op("add"), x, x);
m1.add_return({sum});
}
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
// Test with non-standard strides in allocation
TEST_CASE(realloc_nonstandard_strides)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {3, 2}});
// Allocate with transposed strides {1, 3} but operator expects standard strides
auto alloc =
m1.add_instruction(test_allocate{{migraphx::shape::float_type, {3, 2}, {1, 3}}});
m1.add_instruction(simple_op{{migraphx::shape::float_type, {3, 2}}}, x, alloc);
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {3, 2}});
// After adjust_allocation, should have standard strides
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {3, 2}}});
m2.add_instruction(simple_op{{migraphx::shape::float_type, {3, 2}}}, x, alloc);
}
EXPECT(m1.sort() == m2.sort());
}
// Test with different data types
TEST_CASE(realloc_different_dtype)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::half_type, {4, 4}});
// Allocate with wrong dimensions for half type
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::half_type, {2, 8}}});
m1.add_instruction(simple_op{{migraphx::shape::half_type, {4, 4}}}, x, alloc);
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::half_type, {4, 4}});
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::half_type, {4, 4}}});
m2.add_instruction(simple_op{{migraphx::shape::half_type, {4, 4}}}, x, alloc);
}
EXPECT(m1.sort() == m2.sort());
}
// Test with multiple instructions, only some need reallocation
TEST_CASE(realloc_mixed_instructions)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto y = m1.add_parameter("y", {migraphx::shape::float_type, {3, 2}});
// First op: needs reallocation (allocated {3, 2} but returns {2, 3})
auto alloc1 = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {3, 2}}});
auto r1 = m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, alloc1);
// Second op: no reallocation needed (shapes match)
auto alloc2 = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {3, 2}}});
auto r2 = m1.add_instruction(simple_op{{migraphx::shape::float_type, {3, 2}}}, y, alloc2);
auto sum = m1.add_instruction(migraphx::make_op("add"), r1, r1);
m1.add_return({sum, r2});
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto y = m2.add_parameter("y", {migraphx::shape::float_type, {3, 2}});
// First op: reallocated to correct shape
auto alloc1 = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
auto r1 = m2.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, alloc1);
// Second op: unchanged
auto alloc2 = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {3, 2}}});
auto r2 = m2.add_instruction(simple_op{{migraphx::shape::float_type, {3, 2}}}, y, alloc2);
auto sum = m2.add_instruction(migraphx::make_op("add"), r1, r1);
m2.add_return({sum, r2});
}
EXPECT(m1.sort() == m2.sort());
}
// Test that instructions with no inputs are skipped
TEST_CASE(skip_no_inputs)
{
migraphx::module m1;
{
auto lit =
m1.add_literal(migraphx::literal{{migraphx::shape::float_type, {2, 2}}, {1, 2, 3, 4}});
m1.add_return({lit});
}
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
// Test 3D tensor reallocation
TEST_CASE(realloc_3d_tensor)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3, 4}});
// Allocate wrong 3D shape
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {4, 3, 2}}});
m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3, 4}}}, x, alloc);
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {2, 3, 4}});
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3, 4}}});
m2.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3, 4}}}, x, alloc);
}
EXPECT(m1.sort() == m2.sort());
}
// Test that adjust_allocation inserts a copy when output alias is a parameter with different shape
TEST_CASE(insert_copy_output_param)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
// Output parameter with different shape than what the operator produces
auto out = m1.add_parameter("output", {migraphx::shape::float_type, {3, 2}});
auto r = m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, out);
m1.add_return({r});
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto out = m2.add_parameter("output", {migraphx::shape::float_type, {3, 2}});
// New allocation with correct shape replaces the parameter
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
auto r = m2.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, alloc);
// Copy from result to output parameter
auto c = m2.add_instruction(migraphx::make_op("test::copy"), r, out);
m2.add_return({c});
}
EXPECT(m1.sort() == m2.sort());
}
// Test that copy insertion updates multiple users of the result
TEST_CASE(insert_copy_multiple_users)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto out = m1.add_parameter("output", {migraphx::shape::float_type, {3, 2}});
auto r = m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, out);
// Multiple uses of the result after the instruction
auto sum = m1.add_instruction(migraphx::make_op("add"), r, r);
m1.add_return({sum});
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto out = m2.add_parameter("output", {migraphx::shape::float_type, {3, 2}});
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
auto r = m2.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, alloc);
// Copy from result to output parameter
auto c = m2.add_instruction(migraphx::make_op("test::copy"), r, out);
// Users after the copy should use the copy result
auto sum = m2.add_instruction(migraphx::make_op("add"), c, c);
m2.add_return({sum});
}
EXPECT(m1.sort() == m2.sort());
}
// Test copy insertion with chain of operations using the result
TEST_CASE(insert_copy_chain_of_ops)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto out = m1.add_parameter("output", {migraphx::shape::float_type, {3, 2}});
auto r = m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, out);
// Chain of operations using r
auto neg = m1.add_instruction(migraphx::make_op("neg"), r);
auto relu = m1.add_instruction(migraphx::make_op("relu"), neg);
m1.add_return({relu});
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto out = m2.add_parameter("output", {migraphx::shape::float_type, {3, 2}});
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
auto r = m2.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, alloc);
auto c = m2.add_instruction(migraphx::make_op("test::copy"), r, out);
// Chain uses the copy result
auto neg = m2.add_instruction(migraphx::make_op("neg"), c);
auto relu = m2.add_instruction(migraphx::make_op("relu"), neg);
m2.add_return({relu});
}
EXPECT(m1.sort() == m2.sort());
}
// Test copy insertion with non-standard strides in output param
TEST_CASE(insert_copy_nonstandard_strides)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {3, 2}});
// Output parameter with transposed strides
auto out = m1.add_parameter("output", {migraphx::shape::float_type, {3, 2}, {1, 3}});
auto r = m1.add_instruction(simple_op{{migraphx::shape::float_type, {3, 2}}}, x, out);
m1.add_return({r});
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {3, 2}});
auto out = m2.add_parameter("output", {migraphx::shape::float_type, {3, 2}, {1, 3}});
// Allocate with standard strides
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {3, 2}}});
auto r = m2.add_instruction(simple_op{{migraphx::shape::float_type, {3, 2}}}, x, alloc);
auto c = m2.add_instruction(migraphx::make_op("test::copy"), r, out);
m2.add_return({c});
}
EXPECT(m1.sort() == m2.sort());
}
// Test that copy is not inserted when result is not used after the instruction
TEST_CASE(insert_copy_result_not_used_later)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto y = m1.add_parameter("y", {migraphx::shape::float_type, {2, 3}});
auto out = m1.add_parameter("output", {migraphx::shape::float_type, {3, 2}});
// r is not used after itself - DCE will remove this and the copy
m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, out);
// y is used in return, not r
m1.add_return({y});
}
run_pass(m1);
migraphx::module m2;
{
m2.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto y = m2.add_parameter("y", {migraphx::shape::float_type, {2, 3}});
m2.add_parameter("output", {migraphx::shape::float_type, {3, 2}});
// After DCE, the simple_op, allocate and copy are all removed
m2.add_return({y});
}
EXPECT(m1.sort() == m2.sort());
}
// Test that view operations as output buffer are not modified (shallow alias only traces one level)
// The pass uses shallow=true, so it only looks at the immediate output alias, not through view ops
TEST_CASE(skip_aliased_through_transpose)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {3, 2}});
// Allocate {2, 3} and transpose to {3, 2} - pass won't modify since alias is transpose, not
// allocate
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
auto t =
m1.add_instruction(migraphx::make_op("transpose", {{"permutation", {1, 0}}}), alloc);
m1.add_instruction(simple_op{{migraphx::shape::float_type, {3, 2}}}, x, t);
}
// With shallow=true aliasing, the pass sees transpose as the alias, not the allocate
// Since transpose is not an allocate or parameter, the pass skips this instruction
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
// Test that squeeze as output buffer is skipped (shallow alias)
TEST_CASE(skip_aliased_through_squeeze)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {1, 2, 3}}});
auto sq = m1.add_instruction(migraphx::make_op("squeeze", {{"axes", {0}}}), alloc);
m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, sq);
}
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
// Test that unsqueeze as output buffer is skipped (shallow alias)
TEST_CASE(skip_aliased_through_unsqueeze)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {1, 2, 3}});
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
auto usq = m1.add_instruction(migraphx::make_op("unsqueeze", {{"axes", {0}}}), alloc);
m1.add_instruction(simple_op{{migraphx::shape::float_type, {1, 2, 3}}}, x, usq);
}
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
// Test that slice as output buffer is skipped (shallow alias)
TEST_CASE(skip_aliased_through_slice)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {4, 6}}});
auto sl = m1.add_instruction(
migraphx::make_op("slice", {{"axes", {0, 1}}, {"starts", {0, 0}}, {"ends", {2, 3}}}),
alloc);
m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, sl);
}
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
// Test that transposed parameter as output buffer is skipped (shallow alias)
TEST_CASE(skip_aliased_param_through_transpose)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {3, 2}});
auto out = m1.add_parameter("output", {migraphx::shape::float_type, {2, 3}});
auto t = m1.add_instruction(migraphx::make_op("transpose", {{"permutation", {1, 0}}}), out);
auto r = m1.add_instruction(simple_op{{migraphx::shape::float_type, {3, 2}}}, x, t);
m1.add_return({r});
}
// Shallow alias sees transpose, not the parameter, so pass skips this
migraphx::module m2 = m1;
run_pass(m1);
EXPECT(m1.sort() == m2.sort());
}
TEST_CASE(fill_allocation)
{
migraphx::module m1;
{
auto x = m1.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto alloc = m1.add_instruction(test_allocate{{migraphx::shape::float_type, {3, 2}}});
auto fill = m1.add_instruction(migraphx::make_op("test::fill"), alloc);
m1.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, fill);
}
run_pass(m1);
migraphx::module m2;
{
auto x = m2.add_parameter("x", {migraphx::shape::float_type, {2, 3}});
auto alloc = m2.add_instruction(test_allocate{{migraphx::shape::float_type, {2, 3}}});
auto fill = m2.add_instruction(migraphx::make_op("test::fill"), alloc);
m2.add_instruction(simple_op{{migraphx::shape::float_type, {2, 3}}}, x, fill);
}
EXPECT(m1.sort() == m2.sort());
}
int main(int argc, const char* argv[]) { test::run(argc, argv); }
You can’t perform that action at this time.